134 



NA TURE 



[January 31, 1901 



REFRACTION WITHIN TELESCOPE TUBE. 

 T has been supposed that the difference between the zenith- 

 distance of a star obtained by direct observation and that 

 obtained by observing it reflected in a pool of mercury— as it is 



1 



tttrl 



Fig. I.— a, a, Rays normally refracted by horizontal stratification of air. b, 6', Rays 

 refracted by stratification of air parallel to ;««, m'n', (inverted refraction). ;««, 

 j;i'n', Equal-density surfaces when upper side of tube is cooled. 



not fully explicable as the result of the flexure of the telescope- 

 tube — is partly to be accounted for as the result of abnormal 

 refraction in the neighbourhood of the instrument, 

 owing to varying air-temperatures in the room. It is f 



wilhin the tube of the instrument that these things are 

 to be looked for. 



The air within a room where there is no powerful 

 source of heat and where currents are caused by open 

 shutters, &c. , must be very nearly uniform in tempera- 

 ture ; neither would it prove the contrary to obtain 

 varying readings of the thermometer within the room, * 



such readings being much affected by radiation from the 



walls, &c. On the other hand, the air within the 



telescope-tube must commonly be stagnant, and any 



cause operating to produce differences of temperature 



therein will do so effectually. Now in making any 



observation at considerable zenith distance the upper side 



of the tube is cooled by radiation, while the lower is * 



protected, and the resulting difference of temperature in 



the "metal of the tube is communicated to the air within, 



to an extent depending on the time of exposure. 



In the diagram (Fig. i) showing the telescope in the 



two positions of observing — direct and reflex — the rays 



a, a', are normally refracted and convex upwards, as 



they would be with horizontal stratification of the air. 



At any two points at the same level these rays (supposed 



from the same star) have the same inclination to a 



vertical line ; therefore the zenith distances that would 



be observed with such refraction (angle e, f, g direct and 



e' , f, g' reflex) would be the same — neglecting flexure 



and the difference of latitude between the trough and 



instrument. But this result depends on the supposed 



parallel (viz. horizontal) stratification of the air through- 

 out the course of both rays. When there is a transverse 



gradient of temperature in the tube owing to the cooling 



of its upper side and the contiguous air within, the 



equal-density surfaces are inclined like ni, n, m' , n' — a 



difference of temperature of only two-tenths <5f a degree 



Fahrenheit will make the air at m as dense as it is at n if these 



points differ in level by ten feet — and the stratification of the air, 



being no longer parallel in the two cases, two different zenith- 

 distances will be obtained. The rays within the tube will then 

 be convex downwards, b, b' — " inverted refraction " — and the 

 direct observation will give a result in excess of the reflex one 

 by twice the angle /e h. So if the results were only affected 

 by this process the Reflex-minus-Direct results 

 would be negative. The R - D results obtained 

 in the Greenwich observations are commonly posi- 

 tive by reason of the flexure of the tube, and are 

 reduced in i?iagnitude by inverted refraction. 



In the diagram (Fig. 2) the curve a shows the 

 R - D results that would be obtained if affected by 

 instrumental flexure only ; from the formula 2(0" 'So 

 sine zenith distance). This value o"'8o I take to 

 be nearly the correct horizontal flexure (see below). 

 The curve A shows the mean results obtained for 

 the values of R-D in the years 1892-3-4 for 

 south stars, reflex observation taken first. The 

 curve B shows the same results for north stars. 

 The curve r shows a special series of R - D results 

 obtained in the year 1894, all from south stars, the 

 direct observation being made first. The differ- 

 ences between these curves severally and the flexure 

 curve a are accounted for by the inverted refraction 

 in the tube, and the various values of these differ- 

 ences f( r the three curves are readily explained :— 

 (a) The north-star-curve differs more from the 

 flexure-curve than the south-star-curve does ; the 

 difference in both cases is due to the exposure of 

 the instrument in the position directed to the star 

 (chiefly in the reflex position) ; but in observing a 

 north star the time of exposure is commonly 

 greater than in observing a south star, as the 

 observation in right ascension is made at the same 

 time, and a slow moving polar star requires more 

 time for this purpose. 



(3) From Fig. i it is apparent that it is at the 

 object-glass end of the tube that the inverted re- 

 fraction is most effective in separating the star 

 image h from the position /, where the normal 

 refraction would place it ; and of the two observing positions of 

 the instrument, the direct one chiefly affects the object-glass end 



<$' *#' JJ' .5" 



Fig. 2.— a, Flexure curve, 2(o"-8o sine zenith distance), a, R - D curve, south star?, 

 reflex observation taken first. B, R - D curve, north stars, reflex observation 

 taken first, c, R - D curve, south stars, direct observation taken first. 



by radiation, because in this position it is near the open shutter. 

 In the observations from which the curves A and B are deduced, 



NO. I 63 I. VOL. 63] 



